摘要
目的针对目前大多数焊接缺陷自动特征提取方法存在的准确度较低的问题,研究满足准确度要求的X射线图像中焊接缺陷特征提取方法。方法对图像进行增强去噪预处理后,在初步确定焊缝区域的基础上,根据焊缝图像列灰度值曲线梯度特性,设计基于灰度值梯度的焊缝边界精确提取算法;以提取得到的焊缝精确边界为初始轮廓,提出基于改进DRLSE模型的焊接缺陷特征提取方法。结果基于改进DRLSE模型的焊接缺陷特征提取方法能够有效地提取气孔、夹渣、未熔合和未焊透等缺陷特征,准确率达到94.6%。结论所提方法克服了原始焊缝X射线图像质量较差、背景复杂的问题,能够精确提取焊缝区域边界,并准确地对各种焊接缺陷进行特征提取,具有较强的适应性和实用性。
The work aims to study a feature extraction method for welding defects in X-ray images that meets the accuracy requirements,aiming at the low accuracy of most current automatic feature extraction methods for welding defects.After image enhancement and denoising preprocessing,based on the initial determination of the weld area,and according to the gradient characteristics of the gray value curve of the weld image column,an accurate extraction algorithm of the weld boundary based on the gray value gradient is designed;the precise boundary of the weld is the initial contour,and a feature extraction method of welding defects based on the improved DRLSE model is proposed.The results show that the welding defect feature extraction method based on the improved DRLSE model can effectively extract defect features such as pores,slag inclusions,incomplete fusion and incomplete penetration,and the precision rate reaches 94.6%.The method overcomes the problems of poor image quality and complex background of the original welding seam X-ray image,can accurately extract the boundary of the welding seam region,accurately perform feature extraction on various welding defects,and has strong adaptability and practicality.
作者
章诚
宋燕利
李汉培
高昶霖
左洪洲
ZHANG Cheng;SONG Yan-li;LI Han-pei;GAO Chang-lin;ZUO Hong-zhou(a.Hubei Key Laboratory of Advanced Technology for Automotive Components,Wuhan University of Technology,Wuhan 430070,China;Hubei Collaborative Innovation Center for Automotive Components Technology,Wuhan University of Technology,Wuhan 430070,China;Hubei Research Center for New Energy&Intelligent Connected Vehicle,Wuhan University of Technology,Wuhan 430070,China;Hubei Qixing Automobile Boy Co.,Ltd.,Hubei Suizhou 44300,China)
出处
《精密成形工程》
北大核心
2022年第12期153-159,共7页
Journal of Netshape Forming Engineering
基金
湖北省技术创新专项(2019AAA014)
教育部创新团队发展计划(IRT17R83)
新能源汽车科学与关键技术学科创新引智基地(B17034)。
